INDEX

Explanations

attends to tokens associated with specific categories from tokens that are in related textual contexts

oai_attention-head · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 10-gemmascope-att-16k

Configuration

google/gemma-scope-2b-pt-att/layer_10/width_16k/average_l0_70

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.10.attn.hook_z

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.08

1:0.07

2:0.53

3:0.09

4:0.05

5:0.03

6:0.04

7:0.08

Negative Logits

 autorytatywna

-0.58

Obrázky

-0.51

 disambiguazione

-0.50

GEBURTSDATUM

-0.48

PerformLayout

-0.48

 AppColors

-0.47

 ujednoznacz

-0.46

 &___

-0.45

Portale

-0.43

 <<<<<<<<<<<<<<

-0.43

POSITIVE LOGITS

0.29

0.26

0.24

0.23

さあ

0.23

 fait

0.23

0.22

Activations Density 0.158%

attends to tokens associated with specific categories from tokens that are in related textual contexts

No Comments

No Known Activations